75N99223R0004 PSI BAA Amend 17 NOV.pdf

75N99223R0004, PSI

Broad Agency Announcement (BAA) Precision Surgical Interventions (PSI) Health Science Futures (HSF) Office

75N99223R0004 21 August 2023

Amended 11 September 2023 Amended 17 November 2023

Table of Contents

PART I: OVERVIEW INFORMATION

PART II: FULL TEXT OF ANNOUNCEMENT

1. Funding Opportunity Description

1.1. PROGRAM OVERVIEW

1.2 TECHNICAL APPROACH AND STRUCTURE

1.3 PROGRAM METRICS

1.4 GENERAL REQUIREMENTS

2. Award Information

2.1 GENERAL AWARD INFORMATION

3. Eligibility Information

3.1. ELIGIBLE APPLICANTS

3.2. ORGANIZATIONAL CONFLICTS OF INTEREST (OCI)

4. Application and Submission Information

4.1. ADDRESS TO REQUEST APPLICATION PACKAGE

4.2. CONTENT AND FORM OF APPLICATION SUBMISSION

4.3. FUNDING RESTRICTIONS

4.4. QUESTIONS

5. Application Review Information

5.1. EVALUATION CRITERIA

5.2. REVIEW OF ABSTRACTS AND FULL PROPOSALS

6. Award Administration Information

6.1. SELECTION NOTICES AND NOTIFICATIONS

6.2. ADMINISTRATIVE AND POLICY REQUIREMENTS

6.3. REPORTING

6.4. ELECTRONIC SYSTEMS

7. Agency Contacts

8. Other Information

PART I: OVERVIEW INFORMATION

Federal Agency Name – Advanced Research Projects Agency for Health (ARPA-H), Health Science Futures Office (HSF)

Funding Opportunity Title – Precision Surgical Interventions (PSI) Announcement Type – Initial Announcement Funding Opportunity Number – 75N99223R0004 Assistance Listing Number – 93.384 NAICS: 541714 - Research and Development in Biotechnology (except

Nanobiotechnology)

Dates o Posting Date: August 21, 2023 o Proposers’ Day: September 07, 2023

Proposers’ Day Registration Deadline: August 28th, 2023, 12:00 PM

EDT

o Abstract Due Date and time: September 21st, 2023, 12:00 PM EDT o Proposal Due Date and time: November 20th, 2023, 5:00 PM EST

Concise description of the funding opportunity – Despite recent technological advances, there are no practical solutions for some longstanding challenges faced by surgeons in the operating room. Cancer is oftentimes indistinguishable from normal tissue, and occasionally gets left behind, requiring reoperations. Critical anatomical structures such as nerves, blood vessels, lymph nodes or lymph ducts also look like surrounding tissue, and can be accidentally injured during surgeries, leading to long-term patient morbidity. The Precision Surgical Interventions (PSI) program aims to develop technologies to aid surgeons to complete procedures with high accuracy, dramatically reducing inadvertent errors and reoperation rates. More specifically, PSI seeks to develop intraoperative end-to-end solutions for enhanced visualization of cancerous tissues. These solutions, available at the bedside (imaging the resected specimens), or in vivo (imaging the resection cavity), will allow physicians to visualize with microscopic precision where cancer ends, enabling them to accurately complete all oncological surgeries and reduce reoperations. Secondly, PSI proposes to develop intraoperative systems that enable surgeons to better visualize critical structures in real time. Nerves, blood vessels or other anatomical structures that are not easily visible (e.g., parathyroid glands, urethra, lymph node ducts) will be preserved, thus sparing patients from long-term consequences.

Anticipated individual awards – Multiple awards are anticipated.

Potential award instruments – Cooperative Agreements or Other Transactions (OT) Agency Contact – All inquiries shall be sent to PSI@arpa-h.gov.

PART II: FULL TEXT OF ANNOUNCEMENT

1. Funding Opportunity Description

This publication constitutes a Broad Agency Announcement (BAA) as contemplated in Federal Acquisition Regulation (FAR) 35.016 and Title 2 of the Code of Federal Regulations (CFR) §

200.203 and is in accordance with section 499A of the Public Health Service Act. Advanced Research Projects Agency for Health (ARPA-H) posts this funding opportunity within the framework of a BAA because of its widely recognized use in funding basic and applied research as well as the ability to negotiate multiple award types. Any resultant award negotiations will follow all pertinent laws and regulations.

The mission of ARPA-H is to accelerate better health outcomes for everyone by advancing innovative research that addresses society's most challenging health problems. Awardees will develop groundbreaking new ways to tackle health-related challenges through high potential, high-impact biomedical and health research. ARPA-H seeks proposals to advance strategies that address current surgical challenges and will improve patient outcomes.

Specifically excluded are: 1) proposals that represent an evolutionary or incremental advance in the state of the art 2) partial or incomplete solutions (e.g., contrast agents that label one single type of cancer, microscopic images that are not automatically analyzed and classified, thus requiring a pathologist on staff; imaging and classification approaches that take more than 15 minutes to deliver results) 3) performers unable to address the objectives of the program, 4) proposals directed towards policy changes, traditional education and training, or center coordination and construction of physical infrastructure are outside the scope of the ARPA-H mission.

1.1.PROGRAM OVERVIEW

In modern surgical practice, it is still extremely challenging to distinguish cancer and critical anatomical structures (such as nerves, blood vessels, etc.) from normal surrounding tissue. This inherent lack of contrast results in high reoperation rates and accidental injuries, which can plague patients for years and cost the American healthcare system more than $1 billion per year.

Roughly two million Americans are newly diagnosed with cancer every year. For many, tumor resection is the first-line treatment, and should result in negative (“clean”) margins, i.e., no cancer is left behind. Currently, the gold standard for tumor margin evaluation is a pathological examination of the resected specimen by a board-certified pathologist, usually days after the initial surgery. If the margins are positive, patients generally undergo reoperation. The estimated costs for reoperations due to positive margins after breast cancer surgeries alone are >$500 million/year.

Reoperation is unfortunately not always possible, and patients are offered adjuvant treatment instead (such as chemotherapy or radiation therapy). Survival rates, however, decrease significantly with positive margins; for example, the 5-year survival rate for prostate cancer patients decreases from 71% with negative margins to 42% with positive margins.

The inability to visualize critical anatomical structures also has dramatic consequences for patients undergoing surgery. Unintentional damage or removal of critical structures can cause serious complications both during and after the procedure. Damaged nerves lead to surgically induced neuropathic pain, which occurs in roughly 10 – 50% of patients, with 2 – 10% experiencing severe pain; nicked blood vessels result in bleeding that extends procedures and hospital stays and can threaten the patient’s life; urethra damage can lead to urinary incontinence; parathyroid gland removal can cause hypocalcemia; damage to lymph ducts can cause lymphedema. While some damage to critical structures is both inevitable and planned, many injuries are inadvertent and can be avoided if the surgeon could properly visualize these structures.

The aim of the Precision Surgical Interventions (PSI) program is to catalyze advancements in the surgical field, with the final goals of providing the surgeon with revolutionary tissue visualization and classification tools, thus increasing surgical precision, decreasing reoperations, and improving patient care. PSI seeks to develop technologies that improve surgical outcomes through two technical areas. Technical Area 1 (TA1) aims to develop systems that image tumors intraoperatively at microscopic scales. TA1 requires performers to develop an end-to-end pathology system that operates at the bedside (TA1-A) or in vivo (TA1-B) and classifies margins as positive or negative within 15 minutes, without a pathologist. Technical Area 2 (TA2) aims to develop devices and software to localize and visualize critical anatomical structures (nerves, blood vessels and organs) in 3D during surgery. TA2 requires performers to develop a real-time, end-to-end system that enables visualization of critical structures buried up to 1 cm deep. In accordance with the equity goals of ARPA-H, PSI will require solutions that will be easily accessible to all hospital settings.

1.2 TECHNICAL APPROACH AND STRUCTURE

1.2.1. Technical Areas (TAs)

The PSI program will develop novel intraoperative devices and tools to decrease the reoperation rate in oncological surgeries and the rate of accidental damage to critical structures in all surgeries.

To accomplish this, the PSI Program is focused on two (2) Technical Areas:

Technical Area 1 (TA1): Cancer localization.

Two alternative options for this technical area exist, as described below. Proposers must select one (1) single option.

o TA1-A (In vitro pathology): Development of end-to-end solutions, including devices and software, for intraoperative microscopic imaging of a resected specimen and automated image classification. Use of existing staining agents, or development and optimization of new staining agents, that can be used once the specimen is removed from the body, are also allowed under this aim.

o TA1-B (In vivo pathology): Development of end-to-end solutions, including devices and software, for intraoperative microscopic imaging of the resection cavity and automated image classification. Use of existing, Food and Drug Administration (FDA)-approved contrast agents, as well as development and optimization of new contrast agents, capable of labeling the cancer cells or the cells in the tumor microenvironment (in vivo) are also allowed under this aim.

Technical Area 2 (TA2): Healthy structure localization. Development of devices and software capable of sensing structures of interest at their depth, integrated with surgical tools and/or surgical workflow. Use of existing, FDA-approved contrast agents, as well as development and optimization of new contrast agents capable of labeling the structures of interest in vivo, are also allowed under this aim.

Proposal details: Performers will have the option of submitting proposals that address TA1-A or TA1-B or TA2. If two equally rated proposals address TA1-A and TA1-B, respectively, the one addressing TA1-B will be given preference. A group (e.g., a contrast agent or digital pathology developer) may participate in two proposals but cannot be the prime (contact) proposer on both.

Proposals may only address TA1-A, TA1-B, or TA2, with the single exception of a combined response to TA1-B and TA2, in which a significant component of the technical solution (e.g., microscopy method or contrast agent) addresses both Technical Areas.

Proposals that only provide parts of the solutions above (i.e., just an agent, or just a device, or just a classification algorithm), and not an end-to-end solution, will be deemed non-conforming and rejected without further review.

TA1: Cancer localization

Surgical resection is oftentimes the first line of treatment after cancer diagnosis. Even with thorough preoperative imaging and surgical planning, it is difficult to achieve complete tumor resection, leaving no cancer cells behind, as cancerous tissue cannot be easily distinguished from normal tissue. The surgeon removes a volume of tissue believed to encompass the entire tumor, closes the patient, then sends the resected tissue for pathological examination. Formalin-Fixed Paraffin-Embedded (FFPE) ~4 mm thick slices are obtained from the surgical sample; the surface of these slices is examined under a microscope by a pathologist days later. It is not uncommon for tumor cells to be found at the edge of the surgical specimen, indicating that the tumor was not entirely resected. The patient then usually undergoes a second surgery to remove additional tissue, increasing anxiety, morbidity, and healthcare costs. In some cases, such as prostate cancer, reoperation is not the standard of care, as abdomen reorganization after surgery makes it difficult to locate the effective match to the positive margin identified by the pathologist. Systemic adjuvant therapies are offered instead, but the survival of patients with positive margins after cancer surgery is significantly lower than that of patients with negative margins.

A solution to the above problem is the use of intraoperative frozen section examination during surgery. In this approach, a pathologist is on standby during surgeries for which frozen sections are requested. Verbal communication between the surgeon and pathologist is usually needed to identify the potential regions of interest. Surgery is interrupted, and the specimen is sent for pathology examination; at least 20 minutes are needed for inspection of a single slice by the pathologist, assuming a simple case. Longer time is required if multiple slices need to be inspected, or if it is a complicated case. Freezing tissue introduces artifacts, and the results of frozen section analysis do not always coincide with the results of FFPE analysis, which remains the gold standard.

Although practices and results vary between hospitals, surgeons and pathologists, certain types of surgeries (e.g., for breast cancer) have generally moved away from frozen section analysis. Frozen sections are also not practical in limited resource settings, where a pathologist may not be on call for the duration of the surgery. Even when frozen sections are used, given that only a few slices are examined, FFPE may still indicate additional positive margins, prompting further reoperations.

The vision of the PSI program is that the developments performed under TA1 will reduce positive margins during oncological surgery to no more than 2%. TA1 will require investigators to provide end-to-end solutions, in which either the surgical specimen or the resection cavity are imaged at microscopic resolution. In addition, once images are acquired, an automated analysis will be made, determining if margins contain cancer cells. If such cells exist, their location needs to be identified, allowing the surgeon to further re-excise tissue (during the same initial surgery) if needed. Two implementations of this technical area are envisioned (TA1-A and TA1-B) as described below.

Investigators are expected to select one of the two.

TA1-A: In vitro pathology. Devices and software for intraoperative microscopic imaging of resected specimens and automated image classification will be developed. While it is preferred that the end-to-end solution remain label free for cost and ease-of-use reasons, performers may also propose to use or develop agents to stain the specimen once removed from the body. Agents may include existing, validated staining agents (such as hematoxylin and eosin), those under different stages of development, or brand-new concepts. If new stains are proposed, it is desired that they be cancer-type agnostic, enabling them to be used across the cancer spectrum, and not for one cancer type alone.

The end-to-end solution for automated in vitro digital pathology must meet the following minimum specifications:

It must accommodate samples as large as 10 cm x 10 cm x 10 cm (or an equivalent spherical surface)

Imaging of the entire sample surface is preferred, without sample grossing. If the specimen is grossed, it must provide margin information no sparser than one slice every 4 mm over the volume of the sample (e.g., for a sample of 10 cm on the longest axis, it must be able to image 25 slices). Since margin information alone is needed, readout may be performed only within 1 cm of the edge, should slices be imaged in cross-section.

The samples must be fresh, not frozen.

It must complete the evaluation of the entire sample in less than 10 minutes, including imaging and classification time. An additional 5 minutes for sample preparation may be used.

Imaging resolution must be equal to or better than 0.5 µm. A multiscale approach, in which the majority of the sample is imaged at 1-2 µm, while suspicious regions automatically identified in the lower resolution images are then imaged at 0.5 µm, will also be considered acceptable.

The solution chosen must be generally applicable. To ensure the broad applicability of the approach while keeping program effort reasonable, demonstration in two cancer types is required. For the first cancer, breast, colorectal, head and neck or ovarian cancers should be selected (with the listing order above defining preference order). Any cancer type can be selected for the second validation case. Proposals should include relevant rationale for the choice of the second cancer.

If breast cancer is chosen, additional imaging and classification of sentinel lymph node status is encouraged but not required.

Once images are acquired, classification must be automated (e.g., using a machine learning algorithm). A pathologist must not be required in the operating room.

If the margins are positive, the classification algorithm needs to provide the surgeon with the location of the positive margin, which must have a clear correspondence to the location where the surgeon needs to re-excise.

Final sensitivity, demonstrated in surgical specimens from ≥150 human surgeries, for each of the two cancer types selected, should be ≥98% and specificity ≥95%.

While the primary outcome is resection margin status (2 weeks after primary surgery), secondary outcome measures (listed below) need to be documented.

o Reoperation rate (Time Frame: 2 months after primary surgery). The decision for a reoperation is determined by the surgeon. Data regarding reoperation and reason for reoperation must be recorded.

o Operation time (Time Frame: Immediately after primary surgery). Operation time is defined by time from incision to closure, which will be obtained from anesthesia report.

o Cost effectiveness (Time Frame: 3 months after primary surgery). Data for in-hospital cost including cost for re-excision will be collected. Additional costs due to elongated surgery time and labor costs from the pathology department will also be calculated and included.

o Resection volume (Time Frame: 3 months after primary surgery). Resected volume is calculated from gross specimen measurements of pathology report (volume = width/2 * height/2 * depth/2). When additional resection is performed, resected volumes are reported separately.

If the metrics/number of subjects above are not meaningful for a particular case, proposing teams are expected to provide their own metrics and describe the quantitative improvement those metrics represent over the state-of-the-art. Power analysis calculations are needed to support the proposed metrics.

The ground truth of the pathological read should be set by three pathologists using FFPE.

Since the imaging/validation process is not expected to interfere with standard of care, it is expected that testing of human samples can be completed under IRB approval by the end of the performance period, and not require FDA approvals.

510k/PMA/de novo submission for the device-software combination.

TA1-B: In vivo pathology. Devices and software for intraoperative microscopic imaging of the resection cavity and automated image classification will be developed. While it is preferred that the end-to-end solution remain label free for cost and ease-of-use reasons, performers may also propose to use or develop contrast agent to label cancer macroscopically in vivo. Labeling may come in injectable, oral, paint-on or spray-on form. Contrast agents may include existing, FDA-approved agents (such as Indocyanine Green (ICG), Cytalux or Gleolan), contrast agents already under development, or brand-new concepts. If new agents are proposed, it is desired that they be cancer-type agnostic, targeting general cancer hallmarks (e.g., CD24, B7-H3 markers, annexin, fibroblasts or macrophages, etc.), enabling them to be used across the cancer spectrum, and not for one cancer type alone. Combinations of contrast agents targeting multiple cancer hallmarks (but labeled in a way that enables single querying) are encouraged. Specific solutions (e.g., for prostate cancer), although not preferred, will also be considered, provided a thorough explanation is given as to why a more general solution is not appropriate.

The end-to-end solution for automated in vivo digital pathology must meet the following minimum specifications:

While it is desired that the proposed solution image the entire resection cavity at microscopic resolution, it is understood that motion artifacts and the sheer amount of data may make this impractical. Alternatively, rational/deterministic down-selection from the entire field of view (FOV) to a limited number of regions of interest (ROIs) can be performed. The opinion of a surgeon as to which margin may be positive does not count as a deterministic factor, but selective labeling of a given ROI by a targeted contrast agent does. These ROIs, no less than 2 cm x 2 cm each (or an equivalent circular surface), will then be imaged one at a time.

It must complete the evaluation of each ROI in less than 3 minutes, or of the entire FOV in less than 10 minutes. Imaging time for ROIs smaller than 2 cm x 2 cm will be reduced proportional to the ratio of surface areas.

A support/fixation solution needs to be included in the design, such that an operator is not required to hold on to the imaging probe during the imaging time.

Since the imaging device will be present in the surgical field, sterility needs to be addressed, while not significantly increasing costs.

Imaging resolution must be equal to or better than 0.5 µm.

If a contrast agent/contrast agent cocktail will be used for macroscopic tissue labeling in vivo, >90% of the relevant cancers need to be labeled. For example, if prostate cancer is the main target of the proposal, the agent/cocktail should label 90% of the prostate cancers.

Labeling efficiency for the given agent/cocktail for the second type of cancer studied should be >70%. Alternatively, a different agent/combination may be used for the second cancer type studied, although cross-cancer agents are preferred.

The solution chosen must be generally applicable. To ensure the broad applicability of the approach while keeping program effort reasonable, demonstration in two cancer types is required. For the first cancer, prostate, breast or ovarian cancers should be selected (with the listing order above defining preference order). Any cancer type can be selected for the second validation case. Proposals should include relevant rationale for the choice of the second cancer.

If breast cancer is chosen, additional classification of sentinel lymph node status is encouraged but not required.

If prostate cancer is chosen as the proof-of-concept demonstration, the proposed device needs to be integrated with the surgical robot that has become standard of care for prostate surgeries. Care must be taken such as the proposed solution fits within standard laparoscopic openings.

Once images are acquired, classification needs to be automated (using, e.g., a machine learning algorithm). A pathologist must not be required in the operating room.

If margins are positive, the classification algorithm needs to provide the surgeon with the location of the positive margin.

Final sensitivity should be ≥98% and specificity ≥95%, demonstrated in relevant animal models of the chosen cancer.

If the metrics/number of subjects above are not meaningful for a particular case, proposing teams are expected to provide their own metrics and describe the quantitative improvement those metrics represent over the state-of-the-art. Power analysis calculations are needed to support the proposed metrics.

The ground truth of the pathological read should be set by one pathologist using FFPE.

Since the validation process could interfere with standard of care, it is expected that human testing cannot be completed under IRB approval alone, requiring FDA approvals.

If a contrast agent/contrast agent cocktail is used, to ensure broad clinical availability, it will go through the FDA approval process on its own, not tied to the device and software.

To achieve the goals of the program, performers may propose a variety of technical approaches to classify tumor margins in vitro or in vivo. They may include, but are not limited to:

Contrast agents (optical, ultrasound or radiolabeled) Includes FDA-approved, existing agents already developed but not passed through the FDA approval process, and new agents to be developed through this program.

Given the stringent requirements for labeling efficiency, consider cocktails of agents and partnerships between research groups targeting complementary moieties that are expressed across multiple cancer types (e.g., annexin, fibroblast activation protein and CD24), while ensuring that all targets can be probed simultaneously (e.g., are labeled with the same fluorescing moiety).

Ultrasound or optical microscopy type approaches (Raman, confocal, open top light sheet microscopy, etc.).

Automated image classification approaches. If classification algorithms exist that will meet the performance metrics with minor modifications, they should be employed, as opposed to developing new algorithms.

Proposers must include on their teams as a co-Investigator at least one (1) oncological surgeon with at least five (5) years of experience in resecting the first cancer type chosen. A user experience/ human factor expert also needs to also be included on the team, as a co-investigator or consultant. Consistent interaction between team members will ensure that the solution will fit in the surgical workflow and could be accepted as a standard of care if the project is successful. In addition, within the first three (3) months of performance, interviews need to be conducted with at least 10 other relevant oncological surgeons performing resections of the first chosen cancer type to ensure a thorough understanding exists of what is considered needed/acceptable in the operating room.

Investigators must also provide the following information in the proposal:

Intended in vitro assays and animal models to demonstrate efficacy.

Justification for the number of samples/animals to be used.

Anticipated risks/pitfalls and alternative solutions.

Approximate cost estimate for proposed device.

Cost equation (surgeries saved, costs to the hospital, etc.). An explanation of how the proposed development is expected to be billed/reimbursed (CPT code, insurance pressure, patient pressure, etc.) is also expected.

TA1 metrics (1.3 PROGRAM METRICS) will increase in difficulty and complexity over the course of the PSI program. ARPA-H may request performer data as deemed necessary throughout the program to validate technical progress.

TA2: Critical Structure Localization and Visualization

During interventional procedures, surgeons perform precise mechanical procedures on specific anatomical targets. Many of the tools used, both to perform the procedure and to gain access to the target, are intentionally destructive (scalpels, scissors, etc.); however, surgeons must avoid accidentally damaging critical structures—nerves, blood vessels, ducts, and organs—to prevent long term consequences for the patients’ health. Unfortunately, many critical structures have similar color and texture to surrounding tissue and are difficult to see under standard operating room (OR) lighting, with or without magnification. Laparoscopic surgery suffers from a similar lack of visibility, as the standard white lighting and color video imaging used fail to provide the clarity to distinguish relevant anatomy. Furthermore, many critical structures are buried under other soft tissue. In exposing such buried critical structures, surgeons risk cutting, tearing, or otherwise damaging or destroying them.

Currently, surgical procedures can be informed by preoperative 3D imagery. A radiologist performs and may manually annotate cross sections of a Magnetic Resonance Imaging (MRI), computer tomography (CT), or ultrasound, which the surgeon may be able access for reference during surgery. However, the preoperative imagery does not reflect the status during surgery and orientation of the patient, which is especially of concern when working in highly deformable tissue such as breast. In addition, the structures in the preoperative imagery are not necessarily recognizable using the naked eye (or full spectrum laparoscopic video) in the surgical field. The surgeon thus has the burden of mentally performing 3D transformations, image segmentation, and modality fusion while looking back and forth from the preoperative imagery to the patient.

Fluorescent dyes can be used to help identify critical structures in real time during surgery;

Indocyanine Green (ICG) may be used during angiography procedures, and methylene blue can be used to visualize parathyroid glands. However, the specificity of these dyes is limited, the depth of the structure is not always clear, and the visualization is not integrated or fused into the surgeon’s field of view.

TA2 seeks solutions that impart to the surgeon a real-time, 3D understanding of critical structures in the operating area, thus enabling the clinician to avoid unintentionally damaging critical structures buried under up to 1 cm of other tissue. The solutions must include imaging, visualization, or range-finding components that integrate easily into the surgical workflow.

Preference will be given to solutions that identify multiple critical structures and surrounding anatomy; a solution that only focuses on one or two types of critical structure will also be considered, with preference given to nerves and additional structures (e.g., nerves and urethra, or nerve and lymphatics), or nerves alone.

The critical structure location and visualization solution must meet the following requirements:

The solution must convey 3D understanding to the surgeon such that the surgeon can locate the boundaries of the 3D structure with ±0.5 mm accuracy (up to 2 mm from the surface), ±1 mm (2-5 mm from the surface), and ±2 mm (5-10 mm from the surface)

Most solutions will include at least two components: one component to extract 3D information from the scene (imaging and algorithm), and one to convey that understanding to the surgeon (registration and visualization).

The solution must integrate into the surgical workflow without increasing active operating time by more than 10 minutes. It is acceptable to increase the time devoted to collecting preoperative imagery, or to administer contrast agents before surgery.

The solution must not impede the surgeon’s ability to perform the mechanical aspect of surgery. It must be hands-free, integrated onto an already existing surgical tool, or otherwise out of the way. Wearable headsets must include a setting for un-augmented vision or must allow for easy removal. The sterility of the chosen solution must be addressed.

The initial technology demonstration must be in at least two types of surgeries in which critical structure damage often occurs, such as thoracotomies, mastectomies, prostate cancer removal, or orthopedics.

The solution must be able to visualize critical structures buried up to 1 cm under other tissue.

The solution must update its 3D representation in real time with >= 10 frames per second or 10 Hz update rate

Investigational Device Exemption (IDE) submission If applicable, IND submission

To achieve the goals of the program, performers may propose a variety of technical approaches to locate and visualize critical structures during surgery. These approaches can be separate or combined, and may include but are not limited to:

Fusion of preoperative imagery with real-time intraoperative imagery Computational segmentation of imagery, highlighting nerves, blood vessels, and critical organs Registration and fusion of infrared imagery with visual spectrum imagery in real time Labeling critical structures with contrast agents (oral, intravenous, paint-on or spray-on administration) Agent-free imaging of critical structures using autofluorescence, laser speckle contrast imaging, or other techniques 3D reconstruction from stereo imaging, point clouds, defocus measurements, or other techniques Tool-integrated range finding (e.g., an electrocautery device that measures its distance from nerves and blood vessels) Augmented reality (AR) headset visualization Integrated laparoscopic video visualization

Proposers must include on their teams as a co-Investigator at least one (1) surgeon with at least five (5) years of experience in performing the type of surgery chosen. A user experience/human factor expert also needs to also be included on the team, as a co-investigator or consultant.

Consistent interaction between team members will ensure that the solution will fit in the surgical workflow and could be accepted as a standard of care if the project is successful. Within the first three (3) months of performance, interviews need to be conducted with at least 10 other relevant surgeons to ensure a thorough understanding exists of what is considered needed/acceptable in the operating room. In addition, a user acceptance metric, defined as the overall percent of yes responses to two (2) questions posed to at least five (5) surgeons, needs to be monitored yearly (starting at the end of year 2) and must be provided in the ARPA-H report - with relevant comments (if any). The two questions will be ‘Would you use this solution in your operating room?’ and ‘Do you feel like this solution improves your ability to see critical structures?’.

The following information also needs to be included in the proposal:

Design plan for imaging system, if applicable.

Optimization and validation plan for contrast agent, if applicable.

Technical plan for imaging algorithms, such as image segmentation, real-time 3D reconstruction from 2D images, 3D transformation and projection of high-resolution preoperative imagery onto low-resolution real-time imagery, or multimodal image registration, if applicable.

Design plan for surgical tool integration, if applicable.

Design plan for visualization/display system, if applicable, including an explicit explanation of how the surgeon will perceive depth.

An explanation of how the team will use principles of user experience and human-centered design to create a product that surgeons are able to and want to use.

Plans to conduct user experience and human factor analyses.

Intended tests in phantom tissue, including means to validate accuracy.

Intended tests in animal models, including means to validate accuracy.

Anticipated risks and mitigations.

Cost estimate for the proposed device, including cost equation (complication-related costs avoided, costs to the hospital, etc.). An explanation of how the development is expected to be billed/reimbursed (CPT code, insurance pressure, patient pressure, etc.) is also expected.

TA2 metrics (Section 1.3) will increase in difficulty and complexity over the course of the PSI program. ARPA-H may request performer data as deemed necessary throughout the program to validate technical progress.

1.2.2. Program Structure

The PSI program will be accomplished over two sequential Phases of increasing technical complexity. Decisions to go into Phase II will be determined by the Government based on progress toward achieving Phase I goals. PSI Phases will include programmatic elements to ensure performer success, including a check point between Phases, active and regular US Government stakeholder engagement, equity for disparate patient and market settings for patient/provider buy-in, and utilization of ARPA-H Project Accelerator Transition Innovation Office (PATIO) assets for commercialization (e.g., Expert/Entrepreneur in Residence (XIR/EIR) meetings).

Within the first three (3) months of the program, ARPA-H will organize a Community Symposium, gathering surgeons and hospital administrators from different socio-economic environments, as well as experts in medical devices and reimbursement approaches. At the minimum, the principal investigator from each team with be required to attend, to better understand how to best design the devices to effectively penetrate the market. This symposium may be immediately preceded by an immersion-based kickoff experience in surgical suites for the relevant procedures addressed by the performers. If possible, this immersion experience will be set in a low-resource or underserved community hospital. In addition, ARPA-H will set up advisory boards tailored to the needs of every project, containing members with regulatory expertise, patent/market analysts, individuals who have transitioned similar development towards the product, and practicing surgeons and pathologists across specialties and practice settings (academic, rural and community hospitals, etc.), as needed. They are meant to ensure a smooth transition of the developments from the lab, through FDA, and to the market. Last, extramural resources and labs will serve as independent verification and validation (IV&V) entities throughout the program, to aid in the iterative development of the planned capabilities and validate findings.

Equity Requirements

ARPA-H and PSI are committed to equitable healthcare access irrespective of race, ethnicity, gender/gender identity, sexual orientation, disability, geography, employment, insurance, and socioeconomic status. To that end, PSI will mandate that each performer agree upon and complete the following actions throughout their time in the program:

- Human surgical samples that will be used to test solutions must be reflective of the patient demographics for the indication studied. Performers will be required to perform adequate research to determine what percentages are appropriate.

- The developments of TA1-A will be tested in two hospitals: Most of the human sample testing will be done at the hospital most convenient for the team. For the last two (2) months of testing, the device must be taken to a rural hospital for validation and reliability testing in the last 10 samples. Should this partnership be difficult to set up prior to the submission of the proposal, ARPA-H will help facilitate the connection during the period of performance.

- Performers must thoughtfully design any solution to be compatible for all potential end users. For example, to accommodate surgeons of differing hand sizes, a handheld device could be adjustable. Performers must explain to ARPA-H how their technologies accommodate a diverse range of end users when they submit their prototypes for review.

- At a minimum, the principal investigator of each team must attend a Community Symposium at the end of Month 3 of the period of performance. The Community Symposium will be organized by ARPA-H and will bring together surgeons, clinicians, medical staff, and hospital administrators from diverse medical facilities in the United

States: (1) Academic medical centers, (2) non-academic community medical centers, and

(3) federal government hospitals (e.g., hospitals operated by the Department of Defense, The Department of Human Health and Human Services, or the Veterans Health Administration). The Community Symposium is intended to help performers understand the workflow and cost needs of a diverse range of institutions and clinicians, and to ensure their solutions are broadly applicable and accessible. Performers are encouraged to speak with clinicians and hospital administrators of their own choosing throughout the period of performance. After the symposium, performers should provide a 1-page report of information learned and how the performer will incorporate their learnings into the system design. The report should not detail any confidential information and should be submitted to ARPA-H within four (4) months of starting the project. Only one (1) report is required per performer team, regardless of the number of individuals or institutions the team consists of.

- To prioritize low-cost solutions that can be made more accessible to more end users, the performers must provide a per unit estimated cost analysis of their system that justifies the choice of materials and dimensioning. If a performer chooses a material (e.g., camera, software, computer) that is higher in cost compared to other available materials of the same category, the performer must clearly explain the choice for the higher cost and why a lower cost material is not sufficient. A cost analysis should be provided for the five most expensive items in the system. A preliminary cost analysis should be submitted to ARPA- H within 18 months of funding. A final cost analysis should be submitted before any FDA regulatory filing. Only one set of analyses (primary and final) is required per performer team, regardless of the number of individuals or institutions the team consists of.

Data Sharing Plan

Proposers must agree to openly share deidentified/sanitized data acquired during the period of performance. Any member of the scientific community should have access to the data; registration to a specific repository website is acceptable, but approval needs to be automatic. The specific repository where data will be deposited will be chosen in agreement with the ARPA-H program manager. The proposers will need to present explicit solutions to address the significant data storage and computing challenges presented by the program, with the understanding that the plans and repository may change later in the program.

TA1-A: Phase 1 (24 Months): Proof of concept demonstration

During the 24-month PSI Phase 1, performers will decide whether they will stain the resected tissue. They will perform initial screening and identify the lead compounds, then further optimize these lead compounds. At the same time, they will evaluate potential technologies for in vitro pathology imaging, decide on the lead technology, and develop and optimize the breadboard prototype device that will produce the pathology images.

Goals of PSI Phase 1 (metrics defined in 1.3 PROGRAM METRICS) o Identify workflow and components for the entire in vitro pathology automated readout o If staining agents will be used, begin screening for the appropriate compounds and down-select to three (3) most effective ones each for each cell component to be labeled o Interview 10 relevant surgeons (by month 3) o Attend Community Symposium (month 3) and submit report (by month 4) o Create and submit to ARPA-H initial design history document (by month 12) o Submit Institutional Review Board (IRB) (and Institutional Animal Care and Use committee (IACUC)) applications (as needed) (by month 12) o Develop and optimize breadboard device.

o If staining agents will be used, conduct organoid testing (by month 24) o Prepare prototype for ARPA-H IV&V (by month 23) o Work with PATIO assets to develop commercialization plan (including an engagement plan with Experts in Residence (EIRs) o By month 18: Submit Q-submission meeting package and incorporate feedback;

establish verification and validation study that would be acceptable for FDA (by month 24)

TA1-A: Phase 2 (36 Months): System integration and validation

During the 36-month PSI Phase 2, performers will validate an end-to-end workflow for in vitro automated intraoperative pathology readout for at least two different cancer types.

Goals of PSI Phase 2 (metrics defined in 1.3 PROGRAM METRICS) o If staining agents will be used, down-select to the single most effective one for each cell component/type to be labeled.

o Develop, optimize and validate an integrated, automated device for imaging the surface of the specimen or the volume of the resection in cross section.

o Perform testing in the relevant small animal xenografts and human samples from at least 2 different cancer types.

o Expected performance: 10 min/surgical sample (with an additional 5 min allowed for sample preparation), which can be as large as 10 cm x 10 cm x 10 cm (or the equivalent spherical volume) (by month 60) o Sensitivity should be ≥98% and specificity ≥95%, demonstrated in surgical specimens from >150 human surgeries, for each of the cancer types selected. The ground truth of the pathological read should be set by three pathologists using Formalin-Fixed Paraffin-Embedded (FFPE). (by month 60) o If the metrics/number of subjects above are not meaningful for a particular case, proposing teams are expected to provide their own metrics and describe the quantitative improvement those metrics represent over the state-of-the-art. Power analysis calculations are needed to support the proposed metrics.

o Work with PATIO assets for right-to-practice analysis, patent application, patent landscape analysis, business case, contract research organization (CRO) identification o Next step funding/pathway identified (by month 48) o 510k/PMA/de Novo FDA application (by month 54)

TA1-B: Phase 1 (24 Months): Proof of concept demonstration

During the 24-month PSI Phase 1, performers will decide whether they will use contrast agents for in vivo tissue labeling. They will perform initial screening and identify the lead compounds, then further optimize these lead compounds. At the same time, they will evaluate potential technologies for in vivo pathology imaging, decide on the lead technology, develop and optimize the breadboard prototype device that will produce the in vivo pathology images.

Goals of PSI Phase 1 (metrics defined in Section 1.3) o Identify workflow and components for the entire in vivo pathology automated readout o Interview 10 relevant surgeons (by month 3) o Attend Community Symposium (month 3) and submit report (by month 4) o Create and submit to ARPA-H initial design history document (by month 12) o Submit IACUC application (by month 12) o If contrast agents will be used:

Begin screening for the appropriate compounds and down-select to 3 most effective ones; demonstrate labeling of 70% of relevant cancer types, with a Go/No Go decision on the compound (by month 24)

Conduct INTERACT meeting w/ FDA; finalize verification and validation plan needed for FDA Investigational New Drug (IND) submission. (by month 18)

Conduct basic in vitro/in vivo pharmacology/toxicity tests for the 3 lead compounds with a Go/No Go Decision (for each compound) by month 24.

A single, 100x dose intended to be used in humans will be used (scaled by the ratio of body weights). These tests will include:

In vitro human ether-a-go-go-related gene (hERG) tests indicating no cardiotoxicity

Functional observational battery according to (40 CFR § 798.6050) indicating no neurotoxicity

A liver panel (including, at the minimum, alanine transaminase and aspartate transaminase) indicating no liver toxicity

With aid from PATIO, identify potential GLP and cGMP manufacturing partners, as well as CROs that will perform toxicity, biodistribution, stability studies o ROI identification means finalized (by month 24) o Complete breadboard device; Site preparation time and region of interest (ROI) imaging time: 30min/ROI (by month 24) o Submit Q-submission meeting package and incorporate feedback(by month 15) o Establish verification and validation study (by month 24) o Work with PATIO assets to develop commercialization plan (including an engagement plan with XIR/EIRs) o Prepare prototype for ARPA-H IV&V (by month 23)

TA1-B: Phase 2 (36 Months): System integration and validation

During the 36-month PSI Phase 2, performers will develop an end-to-end workflow for in vivo automated intraoperative pathology readout and test it in at least two different cancer types in two relevant animal models of disease (one rodent and one large animal model)

Goals of PSI Phase 2 (metrics defined in 1.3 PROGRAM METRICS) o Use PATIO assets for right-to-practice analysis, submit patent application, patent landscape analysis, business case, CRO identification (by month 36) o Submit Pre-IDE meeting package and incorporate feedback (by month 36) o If macroscopic cancer labeling agents are part of the foreseen integrated workflow

Having met all criteria of PSI Phase 1, contrast agents must demonstrate 90% cancer labeling in a large animal model with a Go/No Go determination (by month 42)

Submit Pre-IND meeting package and incorporate feedback (by month 36) Secure contract with established partner for producing GLP/cGMP contrast agents and produce agent for 20-100 patients (by month 36) Secure contract with established partner for toxicity and biodistribution/ 2 animal models and complete studies. The toxicity/biodistributions studies agreed to with FDA in prior meetings will be carried out (by month 48).

Secure contract for contrast agent stability studies and complete studies (by month 51)

Submit IND application (by month 54) o Complete integrated device; region of interest (ROI) imaging time: 10 min/ROI (by month 48) o Finalize optimization of classification algorithm; ≥98% sensitivity and ≥95% specificity, measured against ground truth set by one pathologist reading FFPE (by month 54) o Submit IDE application (by month 54) o Imaging time: 3 min/ROI or 10 min/entire cavity (with an additional 5 minutes allowed for cavity preparation) (by month 60) Note: if performers develop a new contrast agent, this agent will undergo regulatory evaluation alone, not in conjunction with the device/software combination.

TA2 Phase 1 (24 Months): Proof of concept demonstration

During the 24-month PSI Phase 1, TA2 performers will develop a prototype device that localizes critical structures and conveys the 3D information to the surgeon. This device should achieve Technology Readiness Level (TRL) 4-5, defined as: subsystems validation and demonstration in a laboratory environment. They will demonstrate the device in phantom tissue, and show that the imaging technology, image processing algorithms, and visualization subsystems work in concert.

If a new contrast agent is developed, they will…